Observations of land surface and snowpack energetics and mass fluxes were made over arctic shrub tundra of varying canopy height and density using radiometers, eddy covariance flux measurements, and snow mass changes from snow surveys of depth and density. Over several years, snow accumulation in the shrubs was found to be consistently higher than in sparse tundra due to greater retention of snowfall by all shrubs and wind redistribution of snowfall to tall shrubs. Where snow accumulation was highest due to snow redistribution, shrubs often became buried by the end of winter. Three classes of shrub-snow interactions were observed: tall shrubs that were exposed over snow, tall shrubs that were bent over and buried by snow, and short shrubs buried by snow. Tall shrubs buried by snow underwent ‘spring-up’ during melt. Though spring-up was episodic for a single shrub, over an area it was a progressive emergence from early to mid melt of vegetation that dramatically altered the radiative and aerodynamic properties of the surface. Short shrubs were exposed more rapidly once snow depth declined below shrub height, usually near the end of melt. Net radiation increased with increasing shrub due to the decreased reflectance of shortwave radiation overwhelming the increased longwave emission from relatively warm and dark shrubs. Net radiation to snow under shrubs was much smaller than that over shrubs, but was greater than that to snow with minimal shrub exposure, in this case the difference was due to downward longwave radiation from the canopy exceeding the effect of attenuated shortwave transmission through the canopy. Because of reduced turbulent transfer under shrub canopies and minimal water vapour contributions from the bare shrub branches, sublimation fluxes declined with increasing shrub exposure. In contrast, sensible heat fluxes to the shrub surface became more negative and those to the underlying snow surface more positive with increasing shrub exposure, because of relatively warm shrub branches, particularly on clear days. From well-exposed tall shrubs, both a large upward sensible heat flow from shrub to atmosphere and a downward flow that contributed substantially to snowmelt were detected. As a result of radiative and turbulent transfer in shrub canopies, melt rates increased with shrub exposure. However, shrub exposure was not a simple function of shrub height or presence, and the transition to shrub-exposed landscape depended on initial snow depth, shrub height, shrub species and cumulative melt, and this in turn controlled the melt energetics for a particular site. As a result of these complex interactions, observations over several years showed that snowmelt rates were generally, but not always, enhanced under shrub canopies in comparison with sparsely vegetated tundra. Copyright © 2006 Crown in the right of Canada. Published by John Wiley & Sons, Ltd.